rec.autos.simulators

I'm a little confused in implementing anti-pitch/drive/squat. The
problem I have currently is that braking the Ferrari in Racer makes it
spin almost always; it's too hard to handle. I believe this is because
of the lack of anti-* modeling in my physics model.
If it helps, I'm reading RCVD on pages 617-619 (suspension geometry).

First of all, I see anti-lift, anti-drive and anti-squat. It seems to
me like anti-squat is the same as anti-lift, only at the rear wheels,
right?

2nd, RCVD talks about inboard and outboard brakes. The result is that
the angles that are used in %anti-x calculates are based on the height
of the instant center from either the ground, or from the center
height of the wheel. What is the difference between the 2 types of
brakes, and what type is typically used? (in for example F1 and Sports
Cars)

If I'm correct, the torque around the center of gravity (My in SAE
terms) that pitches the car is calculated by applying the road force
through or torque arm that goes from the CG height to the (sideview)
instant center height. Right?
In other words, the closer the instant center's height comes to the
center of gravity height, the less pitch moment you will get (when
accelerating/braking).

3rd, there are 2 approaches to handling anti-lift/dive that I
consider: 1 is applying the force (Fx) of the tire to the torque arm
as describe above.
The other one is, if you know the suspension's anti-lift percentage,
is to just calculate a torque from the arm of tire patch (or wheel
center height) to CG height, and instead of applying that, I apply
only AntiLiftPercentage*ThatTorque to the actual torque around the CG.
That *is* the meaning of the percentage, in a way, right?

I think, at this time, specifying the instant centers of the
suspensions (side and front view) is the way to go, and that
calculating the resulting torques will effectively do, without the
explicit need for an anti-dive/lift/squat number. (it's all done using
the side view instant center's height)

As a last thing I don't quite follow, is the statement on page 617,
which says:
"Suspension anti's do not change the steady-state load transfer at the
tire patch."

It seems to me though that by having the car pitch less when having
anti-dive, this results in less weight being put on the front wheels
and more left for the rears.
The problem in Racer is namely that when braking, all weight is
shifted forwards, so the rear wheels give no stabilizing force (as the
carry nearly 0% weight while braking).
I also btw have to do a quick brake-balance parameter, set to the
front by 58% for example (it's now fixed at 50%/50% front/rear), but I
don't think that ultimately decides that my car spins when braking
hard.

Ofcourse, there's always jacking and scrubbing still, but I have a
hunch that their effect is mostly of a 2nd order degree.

Thanks for any clarifications,

Ruud van Gaal, GPL Rank +53.25
Pencil art    : http://www.racesimcentral.net/
Car simulation: http://www.racesimcentral.net/

1. the "bouncing thing" - I know you know about this one already, but I just
thought I'd remind you :p

2. the dials - your revcounter and speedo look really good. Really *really*
good. The nicest instruments I've ever seen in a driving game. Nice one! :0)

3. Start menu shortcuts: the reason all six of these don't work is that they
point to, e.g:
"C:\SIMS\racer.exe" rather than "C:\SIMS\Racer\racer.exe". An easy one to
fix I should think! The "Start In" directories are all correct, just the
target paths are wrong. I'm on Win98 by the way.

4. In "env.ini" there's a discussion of drag due to air - the simple formula
I used at university when some "near enough" numbers were needed was Drag =
0.5 * CD * RHO * A * (V*V)

CD is the drag coefficient, which is around 0.25 to 0.4 for a road-going
saloon car [note that race cars tend to have slightly higher CDs as wings
etc sacrifice "slipperiness" for downforce, sorry but I can't find any
example numbers for race cars].

RHO is the density of air, which is taken to be 1.29 kg per m^3 at standard
temp and pressure.
A is frontal cross-sectional area.
V is velocity relative to air.

5. I get a file called "QLOG.txt" in the Racer directory, which is 30k in
size and consists of many many lines (at least 50, probably more) such as:

Wed Mar 07 22:20:45 (WARN) : [racer/-211417] Can't load material texture
image 'images/POSTA.BMP'

I also get "Wed Mar 07 22:21:05 (WARN) : [racer/-211417] Can't import
'wheel.dof'" and then finally my sidewinder wheel is detected OK. I'll mail
you this text file separately if you want it, incidentally the only file in
my data/images/ directory is "racer512.tga".

That's all I can think of at the moment, keep at it Ruud because I think
you're doing a real nice job of things there!

Hope this helps,

Jim

sorry forgot one - when I press escape to quit Racer I'm returned to my
desktop with commendable speed :) Unfortunately my original resolution and
colour depth doesn't get restored and I have to do it myself in Display
Properties.....

that's it, good luck with it Ruud!

Jim

  I haven't done the anti-xx suspension yet, so can't help you on the specifics
there.  I suspect that if you could take the tire forces and react them through
the linkages themselves, all this stuff would work itself out just fine.  Of
course, I don't know how to do this yet, soooooo..... :0)

  >As a last thing I don't quite follow, is the statement on page 617,

  Pitch won't change total weight transfer.  Anti dive/squat will, however,
change the *portion* of total weight transfer, mainly, the part caused by
spring compression due to pitch.  A car with no suspension will still transfer
the same total weight under a given acceleration.  Total fore/aft weight
transfer is a function of wheelbase, center of mass height, and acceleration
*only.*  If the torques upwards at the front and rear wheels don't equal the
torque caused by longitudinal tire forces (across the ground), the car would
rotate nose first into the ground under braking.  

  >The problem in Racer is namely that when braking, all weight is

  Make sure the weight transfer is correct first, then, get that brake bias in.
 My model will spin around in a real hurry too if 50/50 bias is used.  With the
new driveline model, I can move the bias more rearward than before because it
now fights driveline inertia more accurately, but 50/50 will still cause a spin
in most cases.

  One other thing, the anti-pitch according to "Tune to Win" is usually about
0% in road race cars anyway, while usually remaining below 30% on front engine
vehicles.

  As far as the rest goes, I'm all ears (eyes?) too :0)

Todd Wasson
---
Performance Simulations
Drag Racing and Top Speed Prediction
Software
http://PerformanceSimulations.Com

Locking the fronts first only
keeps you stable if you don't continue pushing harder on the
brakes and go on to also lock the rears ;)
It's my feeling that in a PC sim where you can't count as much
on the driver not doing something stupid (because of the reduced
sensory feedback and loss of fear of death ;)
that the maximum braking torque is almost
as important as the brake bias. They both together determine
when, which axle first, and how easy it is to lock up either
or both front and rears.

A real driver senses yaw acceleration when the car get's squirrely
because the rears are locking or otherwise losing capability.
The PC driver has to wait through 90 or 180 degrees of
additional phase lag before he can sense yaw rate or yaw position
disturbances. By then, it can be too late.

What the suspension geometry does in acceleration is transfer
the forward or backward force to the sprung mass through the suspension
links. If the links don't combine to "point" the total force of the links
onto the sprung mass from the contact patches directly through the CG,
there will be a pitching moment generated. The "Suspension anti's"
change the strength of that moment and also any component of the
forward/aft force that the upward pointing suspension links
bring with them. (The forward or aft force (Fx) has to balance the tire
forces,
but because the links are inclined they also pass Fx tan(theta) upward.
(Simplistically.))

Looking at it from the other direction, from sprung mass down:
more anti-squat means the sprung mass pushes more
directly down on the tires through the links
and less through the springs. With 100% anti-squat the body doesn't
pitch, and the extra force on the rear tires through longitudinal weight
transfer is carried through the suspension links, not the springs.

I happen to be playing around with how this affects a high acceleration
drag race launch at the moment. Although anti-squat doesn't affect the
static weight transfer for a steady acceleration, it does affect the
timing
of the weight transfer in the launch transient because the part that comes

through the springs takes awhile to appear because the sprung mass has
to physically move relative to the unsprunng mass to generate it and
that takes a bit of time.

Like the book says, however, none of this affects the overall amount of
the weight transfer. (These are all internal forces within the overall
multibody system. If you draw a "control volume" around the entire car,
you will still have upward and longitudinal forces at the
tires, and weight down, and forward/aft acceleration at the CG
that all have to balance.)

This is similar to how anti-roll bars don't cause more or less overall
lateral
weight transfer for a given acceleration. They reduce the body roll
angle, but even with zero body roll, the same overall lateral weight
transfer
still occurs. (Because the CG is above the ground, it _has_ to.
One just has to draw and ponder the free body diagram until it blurs ;)

So you can include a lot of anti-dive and still unload the rear tires.
The amount of anti-dive will affect the cars pitch that will affect
the camber change since you will have different suspension travel,
but it means you are transferring more of the weight directly through
the suspension links rather than through the springs and dampers.
And there was a reason you put those springs
and dampers on the  car ;)
But, the weight is still being transferred (normal forces on the tires
changing).
Otherwise the car would flip end for end because the pitch moments
wouldn't balance.

To summarize what I reported above, and as RCVD etc. explains,
this statement is incorrect.

Move the CG down magically, or increase the wheelbase ;),
or just don't brake as hard.
Otherwise you will have to approach 100% front
brake bias as the rears approach being entirely unloaded.
And at that point, brake bias won't help anymore;) ,   you
have lost all the stabilizing effect of the rear tires lateral force
capability anyway, as you say.

Do one of those plots of front brake torque on the vertical axis
and rear brake torque on the lateral axis and plot the torque boundaries
where the fronts and rears will lock.  Gillespie discusses it in his
"Fundamentals of Vehicle Dynamics" in the braking proportioning section.
Be sure to include any friction or motoring torques in the analysis
because if you try to put the brake bias too close, those will
cause you to lock the rears if they aren't included. Also don't forget
to include the effect of the longitudinal load transfer on the tire
maximum
friction coefficients, if you include that. Draw some lines of constant
vehicle
deceleration on the plot, and the straight lines that show the locus
of solutions for several brake bias setttings. Then draw the plot for
full and also for empty tanks. ;)

If you slog through all that, you
will fully understand how your car operates under braking.
Running some steady deceleration tests and plotting the front
and rear tire normal force fractions and comparing it to the simple
longitudinal equations in Gillespie, etc., will also be a good test
(essential, rather) to show that your suspension dynamics are correct.
Given your CG location within the wheelbase and it's height,
you can easily calculate what the weight transfer forward should be.
If it doesn't do this properly, something is wrong. If the basic
weight transfer isn't correct, you are wasting a lot of  work
on ARBs, tire models for friction coefficient change with
normal force, suspension geometry and camber change,
etc., etc.

Anti-dive/squat was tried extensively in the early '60s and basically
abandoned for a reason that wasn't obvious in theory, but which became
quickly onerous in practice: the more you dial in, the more it tends to
lock-up the suspension.  That is, if you rotate the upper A-arm pick-up
points at the front, when the car pitches forward, the anti-dive stiffens up
the front suspension, with an effect similar to hitting the bump ***s.
Very unpleasant.  Spike outside the traction circle.  Ditto the rear: apply
power, and the rear end goes rigid, causing wheelspin.  In those days, there
was plenty of ride height and plenty of suspension travel available (like
off-road racers today), so you didn't really *need* anti-dive/squat; it was
better just to let the chassis pitch (and roll).  Of course, with the
introduction of aero, all that went out the window, and spring rates soared.

--Steve

.......

Bill Shope (one of the original Chrysler Ramchargers drag race team from
the 1960's) did some nice work that models this -- the result turned into
several of the problems for Chapter 17 (pages 32+) in the companion book,
"RCVD Workbook" -- where there is a nice figure that defines the problem.

Our next publishing project is a new version of the Workbook, _with_
_answers_...hopefully this will be out by the end of 2001.  Dragster launch
is not a trivial problem, but Bill worked a closed-form solution that seems
to be correct, at least for the basic situation.

-- Doug

                Milliken Research Associates Inc.

PS. This should be a link to the Workbook, but this version only has
the problems (and some new figures & data), not the answers!
<http://www.sae.org/servlets/productDetail?PROD_TYP=BOOK&PROD_CD=R-212>

--
Matthew V. Jessick         Motorsims

Vehicle Dynamics Engineer  (972)910-8866 Ext.125, Fax: (972)910-8216

Thanks for the feedback. :)

Yep, is solved. The suspension could lock when it hit the max.
Corrected and now also reset with Shift-R/P as well.

Thanks! It's just a speedy TV-Painted dial and a lathe in Max, but it
looks ok. I still find the Nascar dials nicer though. :)

Ah thanks! In Win2K it worked; in my Win95 it didn't indeed. Fixed it
by appending a '\' at the end of the path. Can you believe it! :)

Yep thanks, one of the near-future things; drag and wings.

...

Ah that's nice, the number in SI. :)

No need to worry; a lot of the BMPs are not present actually and
generate these warnings. They show up as red textures in the track.

Phew. :) Well, wheel.dof was a model that just isn't present. I
corrected it by supplying 4 wheel models. Looks much nicer now!

No need to, I know the messages by heart. ;-)

Proceed to the page; v0.39.2 is there NOW! (better braking too)

Thanks,

Ruud van Gaal, GPL Rank +53.25
Pencil art    : http://www.marketgraph.nl/gallery/
Car simulation: http://www.marketgraph.nl/gallery/racer/

Yes, sorry about that one! In Win2K this is ok. But I saw it in Win95
too. It's on the task list...

Ruud van Gaal, GPL Rank +53.25
Pencil art    : http://www.marketgraph.nl/gallery/
Car simulation: http://www.marketgraph.nl/gallery/racer/

It seems it's more a problem of braking torque, CG height and
wheelbase. Well, wheelbase I can't adjust (otherwise the model would
not look correct). I've adjusted the CG hgt and braking torque and
will await messages like 'the thing hardly brakes!'. ;-)

...

Thanks for the explanation. I now see that anti-pitch just changes the
speed/timing at which the pitching occurs, but that's it. And
introduces unresponsiveness in actual cars. I'll keep it at 0 and use
a percentage; it doesn't seem to be that important for the
drivability.

Ruud van Gaal, GPL Rank +53.25
Pencil art    : http://www.marketgraph.nl/gallery/
Car simulation: http://www.marketgraph.nl/gallery/racer/

...

Now, the rears lock first (ofcourse; the front are being pushed hard).
I've tried 100% front brake bias yesterday (or was it the day
before?); that worked quite ok-ish as well!

You're absolutely right. It's better now that I've reduced braking
torque. And reduced the grip coefficients of the tires from 2.4 (from
the original FerP4 file) to 1.7 (more realistic, I'd say). So the
brakes don't grip soooo much that you are thrown forward.
On the top 10 of my task list is generating slip sounds according to
slip ratio, which will make much more realistic slip volumes (it's now
using slipvelocity length).

...

Thanks for the elaborate reply, I think I understand things now
(quickly turned it into a new release just 10 minutes ago).

...

I did 2 of these 3. ;-)
The CG is now at 35cm, which is ok. The wheelbase is locked because I
have a 3D model which must fit. And I reduced braking torque.
AND added a braking_factor to be able to achieve a brake balance. (I
don't have a brake bias since I'm not limited to 4 wheels, but can do
2-8 wheels as well, though I never REALLY tried, so I have to keep
factors apart for each wheel).

...

Phew, I'll check the g's and masses sometime when I have these nice
visual feedback meters which indicate things so I don't have to do too
much by hand. :)

Thanks! It's great to get such good feedback, all of you. :)
The result is on my site (fixed the bugs and some others) as v0.39.2.

Ruud van Gaal, GPL Rank +53.25
Pencil art    : http://www.marketgraph.nl/gallery/
Car simulation: http://www.marketgraph.nl/gallery/racer/

Hehe, I'll bet. But it's nice to see it looks a bit covered now.

...

Sounds very interesting. And if I'm not mistaken, you were working on
a book with a suspension model from Olley? (lots of nice workable
formula's to deal with suspensions)
How's that coming along?

Ah, a shame. :)
I have the workbook, I'll take a look at it.

Thanks,

Ruud van Gaal, GPL Rank +53.25
Pencil art    : http://www.marketgraph.nl/gallery/
Car simulation: http://www.marketgraph.nl/gallery/racer/

How great, I can actually probably leave it out and not notice too
much. :) However, ofcourse, I'll put it in just for the heck of it
(and for the purists and marketing).

Ruud van Gaal, GPL Rank +53.25
Pencil art    : http://www.marketgraph.nl/gallery/
Car simulation: http://www.marketgraph.nl/gallery/racer/

Olley book is nearly ready for the editor/publisher -- the last
photos just arrived and need to be worked in.  I don't know how
long the process will take before release.

-- Doug

                Milliken Research Associates Inc.

....